Yet more "stupid" questions

NancyLebovitz

This is a thread where people can ask questions that they would ordinarily feel embarrassed for not knowing the answer to. The previous thread is at close to 500 comments.

Thanks for the idea. I like the first version of your proposal better than the second, as it risks zero social penalty for wrong guesses.

I'm currently going through Eliezer's long ("intuitive") explanation of Bayes' theorem (the one with the breast cancer and blue-eggs-with-pearls examples), and from what I was able to understand of it, we would need to find out:

Prior: how many of the total men are gay

Conditionals: how many gay men seem to be gay, and how many straight men seem to be gay

... to reach at the posterior (how many men who seem to be gay happen to be gay).

Your proposal sounds useful to solve both conditionals. I guess the main complication is that "to seem to be gay" is terribly difficult to define, and would require endless updates as your life goes through different societies, fads, subcultures, and age groups.

Yea, it might risk social penalties for kidnapping and enslavement, but those seem nowhere as strict. :p

0polymathwannabe13y

OK, I just ran some numbers based on wild guesses. Assuming 10% of all men are gay, and 80% of gay men look gay, and 15% of straight men look gay, my napkin calculation gives about 37% chance that a man who looks gay is actually gay. Doesn't look like any gaydar based on perceived behavior would be too reliable. Of course, if any of my steps was wrong, please let me know.